Background: Moxetumomab pasudotox (Moxe) is a recombinant immunotoxin that targets CD22. Moxe showed high clinical activity in a phase II trial against hairy cell leukemia (Kreitman et al, 2012) and a phase III trial has recently completed enrollment. Moxe also demonstrated objective responses in 30% of children with multiply relapsed acute lymphoblastic leukemia (ALL) (Wayne et al AACR 2014). In a xenograft model with the ALL cell line KOPN-8, Moxe as monotherapy clears the murine bone marrow (BM) from leukemia blasts (Mueller et al, CCR 2016). However, mice eventually relapse and die of progressive disease.Goal: To characterize the ALL cells responsible for relapse and to devise a relapse-preventing treatment.Methods/Results: The ALL cell lines KOPN-8 and REH were transduced with luciferase. The systemic bioluminescence signal in KOPN-8 bearing mice was abrogated after three i.v. bolus-doses of Moxe. Relapse was subsequently detected at various, apparently randomly distributed BM sites which possibly were the origin of systemic relapse. The systemic relapse remained Moxe-sensitive for several treatment cycles, but the bioluminescence signal at the discrete BM sites persisted. After several treatment cycles a Moxe-resistant systemic relapse emerged and mice died of progressive disease as determined by bioluminescence and flow cytometry. The Moxe-resistant cells showed a loss of CD10, and a 2.5-fold decrease in surface CD19 and CD22. CD22 mRNA levels and total CD22 protein by Western Blot were reduced up to 7-fold, indicating a profound reduction of intracellular CD22 in resistant cells. RNA deep sequencing revealed a global transcriptional change including down-regulation of B-cell specific genes suggesting a less mature state of KOPN-8 in the Moxe resistant blasts.We had previsouly demonstrated an epigenetic mechanism of Moxe resistance in vitro (Hu et al, Leuk Res 2013). Because most of the genetically homogeneous KOPN-8 cells growing in murine BM were Moxe sensitive but cells at discrete sites persisted, we hypothesized that locally induced epigenetic changes might partially explain the resistance. We therefore combined Moxe with the DNA-methyltransferase inhibitor 5-Azacytidine (5-AZA). KOPN-8 bearing mice treated with Moxe or 5-AZA alone survived longer than untreated mice, but eventually died from leukemia. However, mice that received the combination of 5-AZA and Moxe were disease free at 4 months after treatment by flow cytometry and histopathology. The combination of 5-AZA and Moxe also improved responses over each agent alone in a second ALL xenograft model with the ALL cell line REH.Conclusion: The relapse of KOPN-8 after treatment with Moxe, possibly from discrete BM sites is consistent with BM-niche induced resistance. This niche-induced resistance is further supported by locally persisting cells after repeated treatment cycles, while blasts that repopulate other BM sites remain rIT-sensitive. That relapse at discrete BM-sites is prevented by 5-AZA suggests that epigenetic modifications contribute to induction of localized resistance in vivo. rIT-resistant KOPN-8 relapse shows a profound reduction in total CD22 which may contribute to resistance to CD22-directed rIT. The marked increase of Moxeefficacy in vivo by the addition of 5-AZA makes this combination a promising approach for further clinical testing in children and adults with ALL. DisclosuresWayne:Kite Pharma: Honoraria, Other: Travel support, Research Funding; Spectrum Pharmaceuticals: Honoraria, Other: Travel Support, Research Funding; Pfizer: Consultancy, Honoraria, Other: Travel Support; Medimmune: Honoraria, Other: Travel Support, Research Funding; NIH: Patents & Royalties. Pastan:NIH: Patents & Royalties: Co-Inventor on NIH Patent.